avoid overwriting existing vai_profile.zip

[rogerngatxilinx]

Keep prior vai_profile.zip archives on successive collects

Summary

Running vaiprofile-collect-data a second time in the same hardware run directory no longer deletes the existing vai_profile.zip. The first archive stays in place, and later collections write vai_profile_1.zip, vai_profile_2.zip, and so on. Collected source files still move into vai_profile_data or the next free vai_profile_data_<n> directory, so both the zip and the moved-file directory now use the same numbered-fallback scheme.

This applies to the shell script and the Windows batch script. Downstream profiling that expects vai_profile.zip still finds the first collection there; later collections are additional archives rather than replacements.

Changes

File	Change
src/mldebug/script/vaiprofile-collect-data	If vai_profile.zip exists, write the next free vai_profile_<n>.zip
src/mldebug/script/vaiprofile-collect-data.bat	Same numbered zip fallback; drop the delete-and-fail path for an existing zip

Before

out="${root}/vai_profile.zip"
rm -f "$out"
zip -j -q "$out" -- "${files[@]}"

The batch script deleted vai_profile.zip and exited with an error if the delete failed.

After

out="${root}/vai_profile.zip"
if [[ -e "$out" ]]; then
  n=1
  while [[ -e "${root}/vai_profile_${n}.zip" ]]; do
    ((n++)) || true
  done
  out="${root}/vai_profile_${n}.zip"
fi
zip -j -q "$out" -- "${files[@]}"

The batch script uses the same loop (vai_profile_%ZN%.zip) before calling :make_zip.

Motivation

vai_profile_data already avoided clobbering a previous collection by stepping to vai_profile_data_<n>. The zip path did the opposite: every run removed vai_profile.zip and wrote a new one, so an earlier archive from the same hardware run directory was lost. A second collect in one directory should keep the prior zip the same way it already keeps the prior data directory.